A Novel Electromagnetic Method for Local Defects Inspection of Wire Rope

Electromagnetic (EM) testing of wire rope is the only effective technology and yields extensive use in the application and manufacturing industry of wire rope. Most EM testing instruments give one-dimensional axial magnetic flux leakage signal, losing the circumferential distribution of defects. Then the identification and classification of the defects largely depend on an operator's experience. Two-dimensional magnetic leakage signal of wire rope is acquired via an inspection prototype with Hall sensor array in this paper, and an adaptive spatial notch filter is designed to eliminate the inherent strand-waveform noise of the wire rope signal. Then a two-dimensional image recognition algorithm is introduced to identify and classify local defects, which includes feature extraction with Karhunen-Loeve (K-L) transformation and defects classification with neural networks. The experimental results show back propagation (BP) network is effective and has a discrimination of 90% for several typical local defects